Crease detection from fingerprint images and its applications in elderly people

Abstract

Conventional algorithms for fingerprint recognition are mainly based on minutiae information. But it is difficult to extract minutiae accurately and robustly for elderly people, and one of the main reasons is that there are many creases on the fingertips of elderly people. In this paper, we study on the detection of creases from fingerprint images, in which we treat the creases as a special kind of texture and design an optimal filter to extract them. We also study the applications of crease detection results to improve the performance of fingerprint recognition in elderly people, which include two aspects. First, it is used to remove the falsely detected minutiae. Second, the creases can be treated as a novel feature for elderly people's fingerprints, which is combined with minutiae feature to improve the performance. Experimental results illustrate the effectiveness of proposed methods.

Introduction

Fingerprint recognition is one of the most popular and reliable biometric techniques for automatic personal identification. It has received more and more attention and been widely used in civilian applications, such as access control and financial security [1], [2]. Since elderly people occupies a rather high proportion (nearly 20 percent of the population) in modern society, it is very important to design a fingerprint recognition system with a satisfying performance for elderly people.

A fingerprint is the pattern of ridges and valleys on the surface of a fingertip. A microscopic feature of the fingerprint is called minutia, which means ridge ending or bifurcation. An ending is a feature where a ridge terminates. A bifurcation is a feature where a ridge splits from a single path to two paths at a Y-junction. In Fig. 1, a fingerprint is depicted, where the ridges are black and the valleys are white. The minutiae, ridge endings and bifurcations, are also shown. Most classical fingerprint verification algorithms [1], [2], [3], [4], [5], [6] take the minutiae, including their coordinates and direction, as the distinctive features to represent the fingerprint in the matching process (see Fig. 1 for the illustrations). Minutiae extraction mainly includes the steps as below: orientation field estimation, ridge extraction or enhancement, ridge thinning and minutiae extraction. Then the minutiae feature is compared with the minutiae template; if the matching score exceeds a predefined threshold, these two fingerprints can be regarded as belonging to a same finger. See Fig. 2 for the flowchart of this kind of algorithms.

However, the quality of fingerprint images may be affected by a number of factors, such as creases, skin dryness, shallow/worn-out ridges, injuries and dirt. This is more serious for elderly people. For these fingerprint images, existing minutiae extraction algorithms are likely to detect spurious minutiae or miss some. As a result, the recognition rate of the fingerprint identification would decrease. Specially, many researches reported that the performance of the minutiae-based algorithms degraded heavily in elderly people's applications [1]. Therefore, it is a challenging problem to improve the performance for elderly people. In this paper, we will address the topic of crease detection from fingerprints and its application in elderly people.

There are always many creases existing in the fingerprints of elderly people, which are a kind of stripes irregularly crossing ridges and valleys in the fingerprints (see Fig. 3 for an example). They come into being because of the aging, manual work, accidents, etc. Some of them are permanent, while others are temporary, i.e., existing for a short term. Both the permanent and temporary creases will affect the orientation estimation, introduce the spurious minutiae and thus decrease the performance of fingerprint identification systems, even using popular preprocessing or post-processing steps (e.g. connecting the breaks by analyzing the orientation filed, or removing adjacent minutiae pairs with opposed directions).

So it is important to study this kind of pattern for fingerprints recognition. If we can detect the creases in advance, we could locate the spurious minutiae caused by creases and remove them. Consequently, the recognition result will be improved. Furthermore, if we can use the crease as a novel kind of feature for accessorial representation to elderly people's fingerprints, it may also result in an improvement of the performance. As we know, there are no other works aiming on this topic except for our earlier work [7], [8].

In this paper, we will study the topic of fingerprint recognition based on crease detection. The contributions of our paper include: first, we use a parameterized rectangle to represent a crease, design an optimal filter as a detector, and employ a multi-channel filtering framework to detect creases in different orientations; second, we propose an algorithm to remove spurious minutiae detected by conventional fingerprint recognition algorithms and thus improve its performance; third, we treat the creases as a novel feature for elderly people's fingerprints, which is combined with minutiae feature to improve the performance. The experimental results on elderly people show that the performance of conventional fingerprint recognition algorithm can be improved by using the proposed algorithms.

The rest of the paper is organized as follows: in Section 2, the problem of crease detection is studied; in Sections 3 and 4, two applications of using crease detection results are introduced, respectively, i.e., removing spurious minutiae and using them as novel features. Experimental results are presented in Section 5. Finally, the conclusions are drawn in Section 6.